Anthropic signs a $10 billion cloud deal with startup Volta to power its Claude AI

Key points

• Anthropic agreed a reported $10 billion cloud-computing deal with AI startup Volta in August 2026.
• The contract runs for six years and covers compute, the raw processing power needed to train and run AI models.
• Volta is partnering with Bitdeer, a crypto-mining company, to build a data centre in Norway with 133 megawatts of capacity.
• Nvidia's Vera Rubin architecture, the chipmaker's newest generation of AI chips, will power the facility.
• The deal follows separate compute agreements Anthropic recently struck with SpaceX and Amazon.

Anthropicing needs enormous electricity and computing muscle to keep Claude, its family of AI assistants, competitive. Bloomberg originally reported the answer: a $10 billion agreement with Volta, a cloud-computing startup founded earlier this year, later confirmed by TechCrunch AI.

What is Volta, and why Norway?

Volta is a new AI cloud provider, renting out computing power over the internet rather than building AI models itself. It's part of Nvidia's Cloud Partner programme, a consortium of data-centre operators that build their infrastructure around Nvidia's specialist AI chips.

For this deal, Volta has teamed up with Bitdeer, a company best known for mining cryptocurrency, to build a purpose-built data centre in Norway delivering 133 megawatts of capacity. Norway's attraction is straightforward: the country runs almost entirely on hydroelectric power, keeping electricity cheap and relatively clean.

We first covered Volta on 4 August 2026, the same day this deal surfaced. The broader scramble for AI computing capacity has been a recurring theme in our reporting; our 30 July piece on idle GPUs noted that expensive hardware often sits underused even as companies race to lock in more of it.

What chips will it run on?

The centre will use Nvidia's Vera Rubin architecture, the chipmaker's current leading design for AI processing. GPUs (the specialised chips that handle the heavy number-crunching AI requires) are in short supply across the industry. Locking in six years of capacity on next-generation hardware is the real prize for Anthropic.

Why is Anthropic doing so many of these deals?

Competition, simply. Building frontier AI models, the most powerful kind at the cutting edge of the field, requires staggering amounts of compute. Anthropic has signed similar agreements with Amazon and SpaceX in recent months, spreading supply across several providers rather than depending on one.

For everyday Claude users, more compute generally means faster responses and more capable models over time.

Common questions

Does this affect how much Claude costs to use?

Not directly, at least not yet. Compute costs drive AI pricing, but Anthropic hasn't announced any changes tied to this deal. Securing more plentiful compute could help keep costs down eventually.

What is the free tier situation for Claude?

Claude still offers a free tier at claude.ai with daily usage limits, alongside a paid plan. None of that changes with this infrastructure news.

Should I be worried about where my data goes?

Anthropicing's privacy policy covers how it handles user data. The Norway data centre will process Anthropic's compute jobs, which is separate from where your personal conversations are stored. Worth a read if you use Claude for anything sensitive.